Matthew Richards wrote:
Apologies in advance for the long post, and if this question is being posted in
the wrong place. It is rather kickstart oriented but I think this is the right
place for it.
Short version:
In a kickstart install, can the install method (specifically 'harddrive') be
contained within a file that is referenced via a %include option in the main
kickstart file?
Long Version:
I am re-spinning CentOS 5.2 (Anaconda 11.1.2.113-1) with the aim of making it
install unattended from a USB flash drive.
The USB drive is identified differently depending upon the hardware
configuration of the system with which it is booted. Generally speaking if the
system has only IDE disks then the USB drive becomes /dev/sda. If the system
has a SCSI or SATA disk then the USB drive becomes /dev/sdb, etc.
The problem I face is that Anaconda sees the USB drive as a viable install
target and bootloader option. To prevent Anaconda from installing to the USB
drive I am using the harddrive, ignoredisk and bootloader kickstart options,
e.g.
harddrive --partition=/dev/sda1 --dir=/iso
ignoredisk --drives=sda
bootloader --driveorder=hda,hdb,hdc,hdd,sda,sdb,sdc,sdd --location=mbr
(Ok, I know the configuration of the bootloader option is ugly but it meets my
requirements for now and my respun distro will probably never be installed on a
system with more than 3 local disks)
I cannot hard code these kickstart options because the USB drive identity
changes with each different hardware configuration, so I have used a similar
approach to the dynamic partitioning example given in the %pre section of the
kickstart doco. That is, I am using the %pre section to probe the install
target and identify the USB drive and then generate a kickstart file-part that
contains the appropriate harddrive, ignoredisk and bootloader options as
outlined above. The kickstart-part is then included in the main kickstart file
via the %include option.
It seems that when the install method is not in the main kickstart file that
Anaconda prompts for the install method. I believe this is because it does not
run the scripts in the %pre section until after it has loaded the stage2 image,
for which it needs to know the install method which is why it prompts the user
and puts and end to the whole unattended experience. My kickstart file looks
something like this:
%include /tmp/ks-part
%include /tmp/ks-common # this has all of my other kickstart options in it.
%pre
# shell script to identify USB drive and echo 'harddrive' and 'ignoredisk'
options to /tmp/ks-part
Can anyone suggest a way around this or an alternative solution please?
Two that have come to mind while writing this post are (1) use custom udev
rules added via the initrd.img file to ensure that usb drive is always mounted
as /dev/sdX and then hard code this setting into the kickstart file, (2) use a
custom anaconda installer update (e.g. kickstart.py) to perform the dynamic USB
drive identification and exclusion, instead of using the %pre section of
kickstart.
Thanks.
Matt
You can try labeling the usb drive, then use that label to identify the
usb drive at the boot prompt with ks=LABEL="name":/ks.cfg. and
method=hd:LABEL="name"/<path>
Not sure if your version of anaconda will support that syntax.
Jerry
_______________________________________________
Anaconda-devel-list mailing list
Anaconda-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/anaconda-devel-list
